arXiv:1310.5654v2 [astro-ph.EP] 8 Feb 2018 o.Nt .Ato.Soc. Astron. R. Not. Mon. .M .Smith, S. M. A. n .G West G. R. and .Pepe, F. .Gillon, M. .R Anderson, R. D. binary G4+K3 a the of transits primary WASP-70Ab evolved and dwarfs K active transit WASP-84b WASP-6 planets: sub-Jupiter-mass newly-discovered Three cetd21 uut2.Rcie 04Ags 8 norigina in 18; August 2014 Received 22. August 2014 Accepted

sdi hswr r vial tteCSvaanony- via CDS the at available are work this in used p under telescope † 3.6-m ESO the 89.C-0151. on spectrograph HARPS the Euler-Swis EulerCa and 1.2-m photometer, the on TRAPPIST mounted 60-cm both spectrograph, the CORALIE PL12B13 Silla: program La under at Telescope located Liverpool 2-m the on instrument camera survey photometric Palma) (La SuperWASP-North ⋆ 1 3 2 5 4 7 6 8 9 c srpyisGop el nvriy tfodhr T 5 ST5 Staffordshire University, Keele Group, Astrophysics ntttdAtohsqee eG´eophysique, Universit´e de et d’Astrophysique Institut UA colo hsc n srnm,Uiest fS.And St. of University Astronomy, and Physics of School SUPA, bevtied e`v,Uiested e`v,5 Chemin Gen`eve, Gen`eve, 51 Universit´e de de Observatoire eateto hsc,Uiest fWrik oetyCV4 Coventry Warwick, of University Physics, of Department sa etnGopo eecps praod oro 2,E 321, Correos de Apartado Telescopes, of Group Newton Isaac aeds aoaoy hmo vne abig,C30H CB3 Cambridge, Avenue, Thomson J J Laboratory, Cavendish .Cpriu srnmclCnr,Pls cdm fScien of Academy Polish Centre, Astronomical Copernicus N. eateto hsc,adKviIsiuefrAstrophysic for Institute Kavli and Physics, of Department h htmti iesre n ailvlct data radial-velocity and time-series photometric The ae nosrain aewt h APSuh(ot Afric (South WASP-South the with made observations on Based 03RAS 2013 5 .Pollacco, D. 3 .G Y. 445 4 1 1 mzMqe Chew, Maqueo omez , ´ 1412 21)Pitd1 coe 08(NL (MN 2018 October 17 Printed (2014) 1114–1129 , 8 ‡ .Southworth, J. .ClirCameron, Collier A. 4 (0.26 4,ec fwihobt rgtsa ( bright WASP-6 a orbits transiting which the of each of 84b, discovery the report We ABSTRACT negigms-osa aeof rate a at mass-loss undergoing 60 X-rays detected h vprto aeetmtdfrH 048 n D189733b HD and 209458b Lyman- HD anomalously-large ited for estimated rate evaporation the lnt(0.59 planet 0Gr 4K iay ihasprto f3 of separation a with binary, G4+K3 10-Gyr, AP7A lnt n aelts niiul WASP-84b. individual: satellites: and WASP-69b planets individual: – satellites: WASP-70Ab and planets – velocities lnt(0.69 planet e words: Key by unaffected resid essentially between were fit parameters a system the with The reduced pre-whitening span. of this pre-whi method found We we oft-used WASP-84, series. the W harmonic and WASP-69 low-order date, a to using active ground the the For from period. discovered planets transiting the .Queloz, D. M Jup 1.06 , M M eL`g,Ale u6Aˆt 7 a.BC iee1 Belgium Li`ege 1, B5C, Bat. Aoˆut, 17, 6 Li`ege, All´ee du de lnt n aelts eeto ehius photometr techniques: – detection satellites: and planets Jup Jup G UK. BG, 1 n pc eerh ascuet nttt fTechnolog of Institute Massachusetts Research, Space and s e alets 20Suen,Switzerland Sauverny, 1290 Maillettes, des 5 R .H .J Triaud, J. M. H. A. 1.16 , 0.94 , ,teRISE the s, e,Bryk 8 076 asw Poland Warsaw, 00-716, 18, Bartycka ces, , telescope, s om21 coe 21 October 2013 form l A,UK 7AL, Jup 6 es ot ag,Ff Y69S UK 9SS, KY16 Fife Haugh, North rews, n the and m ± 370SnaCu el am,Tnrf,Spain Tenerife, Palma, la de Cruz Santa -38700 n,all and, , .S D. na388dpro ruda active, an around period 3.868-d a in ) 27 ,UK E, rogram 4 )and a) R R .Hellier, C. ′′ 2 Jup Jup rmWS-9 ftesa stesuc hntepae ol b could planet the then source the is star the If WASP-69. from .Delrez, L. egransan, ´ na373dobtaon h rmr faspatially-resolv a of primary the around orbit 3.713-d a in ) na .2- ri ruda active, an around orbit 8.523-d an in ) α osfpt dacusrsgf 107.2.)o via or (130.79.128.5) cdsarc.u-strasbg.fr to ‡ ftp http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/MNRAS/445 mous affecte not are surveys Transit Desider or 2008). 2003; al. an al. et Hu´elamo by et Santos induced 2001; 2004; al. motion et reflex Queloz (e.g. or body source activity ing the stellar whether is determining signal in RV difficulty inherent the star of inactive on focus to tend surveys (RV) Radial-velocity INTRODUCTION 1 ⋆ bopindrn rni.WS-0bi sub-Jupiter-mas a is WASP-70Ab transit. during absorption ∼ -al [email protected] E-mail: † 10 1 12 V A 5 3 .Jehin, E. T s g E ∼ .Skillen, I. .P Doyle, P. A. tl l 22 doi:10.1093/mnras/stu1737 v2.2) file style X − 5 0.WS-9 sabotdStr-asplanet Saturn-mass bloated a is WASP-69b 10). 1 ′′ . , hsi – reso antd ihrthan higher magnitude of orders 1–2 is This . 9 ( 3 .D Turner, D. O. > 0 U.WS-4 sasub-Jupiter-mass a is WASP-84b AU). 800 3 .Lendl, M. 7 lnt n aelts individual: satellites: and planets – .Smalley, B. pre-whitening. 1 ∼ a ailvlct n bisector and velocity radial ual S-4 a h third-longest the has ASP-84b .Faedi, F. ,Cmrde A019 USA. 02139, MA Cambridge, y, eiulsatrmr hndid than more scatter residual b AP7A n WASP- and WASP-70Ab 9b, -y,mdKdaf ROSAT dwarf. mid-K 1-Gyr, oho hc aeexhib- have which of both , ∼ ee h ailvelocities radial the tened 1 -y,eryKdaf Of dwarf. early-K 1-Gyr, 5 .Udry S. c–tcnqe:radial techniques: – ic .F .Maxted, L. F. P. 4 /1114 1 b& 9b .Fumel, A. 5 because s d 9– ed, tal. et a fan of by d bit- 3 e s 1 2 D. R. Anderson et al. this ambiguity as stellar activity does not produce transit-like fea- each season of data from each camera separately and we perform tures in lightcurves. The separation of the RV contributions due combined searches on all WASP data available on each star, allow- to reflex motion from those due to activity can prove simple ing for photometric offsets between datasets. This results in early or unnecessary in the case of short-period, giant transiting plan- detection, a useful check for spurious signals and increased sen- ets (e.g. Maxted et al. 2011; Anderson et al. 2012), but it is non- sitivity to shallow transits and long periods. We routinely correct trivial for lower mass planets such as the super-Earth CoRoT- WASP data for systematic effects using a combination of SYSREM 7b (Queloz et al. 2009; Lanza et al. 2010; Pont, Aigrain & Zucker and TFA (Tamuz, Mazeh & Zucker 2005; Kov´acs, Bakos & Noyes 2011; Ferraz-Mello et al. 2011; Hatzes et al. 2011). 2005). TFA is more effective at e.g. removing sinusoidal modu- Transiting planets orbiting one component of a multiple sys- lation, as can result from a non-axisymmetric distribution of star tem have been known for some time (e.g. WASP-8; Queloz et al. spots, which can otherwise swamp the transit signal. However, 2010) and the Kepler mission recently found circumbinary tran- TFA tends to suppress transits whereas SYSREM does not. There- siting planets (e.g. Kepler-16 and Kepler-47; Doyle et al. 2011; fore we use SYSREM-detrended WASP lightcurves, corrected for Orosz et al. 2012). For those systems with a spatially-resolved modulation as necessary, in system characterisation (Sections 4 and secondary of similar brightness to the primary (e.g. WASP-77; 5). Maxted et al. 2013), the secondary can be used as the reference We detected periodic dimmings in the WASP lightcurves of source in ground-based observations of atmospheres. WASP-69, -70 and -84 with periods of 3.868 d, 3.713 d and 8.523 d, Without a suitable reference such observations tend to either fail respectively. We identified WASP-69b and WASP-70Ab as candi- to reach the required precision or give ambiguous results (e.g. dates based on data from the 2008 season alone, whereas data from Swain et al. 2010; Mandell et al. 2011; Angerhausen & Krabbe two seasons (2009 and 2010) were required to pick up the longer- 2011). period WASP-84b. This highlights the importance of multi- The longitudinal coverage of ground-based transit surveys coverage for sensitivity to longer-period systems. The number of such as HAT-Net and HAT-South makes them relatively sensitive to full/partial transits observed of WASP-69b was 2/2 in 2008, 2/5 longer-period planets (Bakos et al. 2004, 2013). For example, in the in 2009 and 2/4 in 2010. The figures for WASP-70Ab are 4/2 in discovery of HAT-P-15b, the planet with the longest period (10.9 d) 2008, 2/5 in 2009 and 2/2 in 2010. The figures for WASP-84b are of those found by ground-based transit surveys, transits were ob- 1/1 in 2009, 1/2 in 2010 and 2/0 in 2011. In the top panels of Fig- served asynchronously from Arizona and Hawai’i (Kov´acs et al. ures 1, 2 and 3 we plot all available WASP data, phase-folded on 2010). By combining data from multiple seasons, surveys such as the best-fitting periods (Section 5) and corrected for systematics SuperWASP can increase their sensitivity to longer periods without with SYSREM; the lightcurves of WASP-69 and WASP-84 have additional facility construction costs (Pollacco et al. 2006). been corrected for rotation modulation (Section 4). Coherent struc- We report here the discovery of three exoplanet systems, each ture is apparent in the lightcurve of WASP-84 with a timescale of comprising of a giant planet transiting a bright host star (V 10). 1 day. This is probably due to an imperfect subtraction of the rota- ∼ ∼ WASP-69b is a bloated Saturn-mass planet in a 3.868-d period tional modulation signal and it does not appear to affect the transit around an active mid-K dwarf. WASP-70Ab is a sub-Jupiter-mass (see the second panel of Figure 3). Though the structure is far re- planet in a 3.713-d orbit around the primary of a spatially-resolved duced in a lightcurve detrended with SYSREM+TFA (not shown), G4+K3 binary. WASP-84b is a sub-Jupiter-mass planet in an 8.523- we opted not to use it as the transit was also suppressed (as was d orbit around an active early-K dwarf. In Section 2 we report the evident from a combined analyis including the RISE lightcurves). photometric and spectroscopic observations leading to the identifi- We obtained spectra of each star using the CORALIE spectro- cation, confirmation and characterisation of the exoplanet systems. graph mounted on the Euler-Swiss 1.2-m telescope. Two spectra of In Section 3 we present spectral analyses of the host stars and, in WASP-84 from 2011 Dec 27 and 2012 Jan 2 were discarded as they Section 4, searches of the stellar lightcurves for activity-rotation were taken through cloud. We also obtained five spectra of WASP- induced modulation. In Section 5 we detail the derivation of the 70A with the HARPS spectrograph mounted on the ESO 3.6-m systems’ parameters from combined analyses. We present details telescope. As the diameter of the HARPS fibre is half that of the of each system in Sections 6, 7 and 8 and, finally, we summarise CORALIE fibre (1′′ cf. 2′′) this is a useful check for contamination our findings in Section 9. of the CORALIE spectra by WASP-70B, located 3′′ away. Radial- velocity (RV) measurements were computed by∼ weighted cross- correlation (Baranne et al. 1996; Pepe et al. 2005) with a numerical G2-spectral template for WASP-70 and with a K5-spectral template 2 OBSERVATIONS for both WASP-69 and WASP-84 (Table 3). RV variations were de- The WASP (Wide Angle Search for Planets) photometric survey tected with periods similar to those found from the WASP photom- (Pollacco et al. 2006) monitors bright stars (V = 8–15) using two etry and with semi-amplitudes consistent with planetary-mass com- eight-camera arrays, each with a field of view of 450 deg2. Each panions. The RVs are plotted, phased on the transit ephemerides, in array observes up to eight pointings per night with a cadence of 5– the third panel of Figures 1, 2 and 3. 10 min, and each pointing is followed for around five months per For each star, we tested the hypothesis that the RV variations season. The WASP-South station (Hellier et al. 2011) is hosted by are due to spectral-line distortions caused by a blended eclipsing the South African Astronomical Observatory and the SuperWASP- binary or by performing a line-bisector analysis of the North station (Faedi et al. 2011) is hosted by the Isaac Newton cross-correlation functions (Queloz et al. 2001). The lack of corre- Group in the Observatorio del Roque de Los Muchachos on La lation between bisector span and RV supports our conclusion that Palma. the periodic dimming and RV variation of each system are instead The WASP data were processed and searched for tran- caused by a transiting planet (fourth panel of Figures 1, 2 and 3). sit signals as described in Collier Cameron et al. (2006) and We performed high-quality transit observations to refine the the candidate selection process was performed as described in systems’ parameters using the 0.6-m TRAPPIST robotic telescope Collier Cameron et al. (2007). We perform a search for transits on (Gillon et al. 2011; Jehin et al. 2011), EulerCam mounted on the

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 3

1.2-m Swiss Euler telescope (Lendl et al. 2012), and the RISE cam- 1.02 era mounted on the 2-m Liverpool Telescope (Steele et al. 2004, 2008). Lightcurves, extracted from the images using standard aper- 1 ture photometry, are displayed in the second panel of Figures 1, 2 and 3 and the data are provided in Table 4. The RISE camera was 0.98 specifically designed to obtain high-precision, high-cadence transit Relative flux lightcurves and the low scatter of the WASP-84b transits demon- 0.4 0.6 0.8 1 1.2 1.4 1.6 strate the instrument’s aptness (see Figure 3). A summary of observations is given in Table 1 and the data 1 are plotted in Figures 1, 2 and 3. The WASP photometry, the radial- velocity measurements and the high signal-to-noise transit photom- etry are provided in tables online at the CDS; a guide to their con- WASP tent and format is given in Tables 2, 3 and 4. 0.98

0.96 3 STELLAR PARAMETERS FROM SPECTRA EulerCam The CORALIE spectra were co-added for each star to produce Relative flux high-SNR spectra for analysis using the methods of Gillon et al. 0.94 (2009) and Doyle et al. (2013). We obtained an initial estimate of (Teff) from the Hα line, then refined the de- termination using excitation balance of the Fe I abundance. The TRAPPIST 0.92 surface gravity (log g ) was determined from the Ca I lines at 6162 ∗ and 6439 (Bruntt et al. 2010b), along with the Na I D and Mg I b lines. The elemental abundances were determined from equivalent- 0.98 0.99 1 1.01 1.02 60 width measurements of several clean and unblended lines. A value for microturbulence (ξt) was determined from Fe I using the method of Magain (1984). The quoted error estimates include that given by 40 the uncertainties in Teff, log g and ξt, as well as the scatter due to -1 measurement and atomic data∗ uncertainties. We determined the projected velocity (v sin I) 20 by fitting the profiles of several unblended Fe I lines. We took 1 macroturbulence values of 2.3 0.3 km s− for WASP-70A and 0 1 ± 1.2 0.3 km s− for WASP-84 from the tabulation of Bruntt et al. ± (2010a). For WASP-69 and WASP-70B we assumed macroturbu- lence to be zero, since for mid-K stars it is expected to be lower than -20 that of thermal broadening (Gray 2008). An instrumental FWHM of 0.11 0.01 was determined from the telluric lines around 6300. Relative / m s -40 The± results of the spectral analysis are given in Table 5 and further details of each star are given in Sections 6, 7 and 8. -60 0.4 0.6 0.8 1 1.2 1.4 1.6 -1 4 STELLAR ROTATION FROM LIGHTCURVE 40 MODULATION 0 We analysed the WASP lightcurves of each star to determine whether they show periodic modulation due to the combination -40 of magnetic activity and stellar rotation. We used the sine-wave Bis. span / m s fitting method described in Maxted et al. (2011) to calculate pe- 0.4 0.6 0.8 1 1.2 1.4 1.6 riodograms such as those shown in Figures 4 and 5. The false Orbital phase alarm probability levels shown in these figures are calculated using Figure 1. WASP-69b discovery data. Top panel: WASP lightcurve folded on a boot-strap Monte Carlo method also described in Maxted et al. the transit ephemeris and binned in phase with a bin-width, ∆φ, equivalent (2011). to two minutes. Second panel: Transit lightcurves from facilities as labelled, Variability due to star spots is not expected to be coherent on offset for clarity and binned with ∆φ = 2 minutes. The best-fitting transit long timescales as a consequence of the finite lifetime of star-spots model is superimposed. Third panel: The pre-whitened CORALIE radial and differential rotation in the so we analysed each velocities with the best-fitting circular Keplerian orbit model. Bottom panel: season of data separately. We also analysed the data from each The absence of any correlation between bisector span and radial velocity WASP camera separately as the data quality can vary between cam- excludes transit mimics. eras. We removed the transit signal from the data prior to calculat- ing the periodograms by subtracting a simple transit model from the lightcurve. We then calculated periodograms over 4 096 uniformly 1 spaced frequencies from 0 to 1.5 cycles day− .

c 2013 RAS, MNRAS 000, 1–17 4 D. R. Anderson et al.

Table 1. Summary of observations

Facility Date Nobs Texp Filter (s)

WASP-69: WASP-South 2008 Jun–2010 Oct 29 800 30 Broad (400–700 nm) SuperWASP-North 2008 Jun–2009 Oct 22 600 30 Broad (400–700 nm) Euler/CORALIE 2009 Aug–2011 Sep 22 1800 Spectroscopy Euler/EulerCam 2011 Nov 10 367 30 Gunn-r ∼ TRAPPIST 2012 May 21 873 10 z ∼ ′ WASP-70 A+B: WASP-South 2008 Jun–2009 Oct 12 700 30 Broad (400–700 nm) SuperWASP-North 2008 Sep–2010 Oct 14 000 30 Broad (400–700 nm) A: Euler/CORALIE 2009 Jul–2011 Oct 21 1800 Spectroscopy A: ESO-3.6m/HARPS 2012 Jun 26–2012 Jul 07 5 600–1800 Spectroscopy B: Euler/CORALIE 2009 Sep 15–22 4 1800 Spectroscopy Euler/EulerCam 2011 Sep 20 466 40 Gunn-r ∼ TRAPPIST 2011 Sep 20 1334 12 I + z ∼ ′ WASP-84: WASP-South 2009 Jan–2011 Apr 14 100 30 Broad (400–700 nm) SuperWASP-North 2009 Dec–2011 Mar 8 700 30 Broad (400–700 nm) Euler/CORALIE 2011 Dec–2012 Mar 20 1800 Spectroscopy TRAPPIST 2012 Mar 01 557 25 z ∼ ′ LT/RISE 2013 Jan 01 4322 4 V + R LT/RISE 2013 Jan 18 1943 8 V + R

Table 2. WASP photometry

Set Star Field Season Camera HJD(UTC) Mag., M σM 2450000 − (day)

1 WASP-69 SW2045 0345 2008 223 4622.483160 9.9086 0.0135 − 1 WASP-69 SW2045 0345 2008 223 4622.483588 9.8977 0.0139 − ... 10 WASP-70 SW2114 1205 2008 221 4622.483391 11.2161 0.0096 − 10 WASP-70 SW2114 1205 2008 221 4622.483831 11.2098 0.0101 − ... 19 WASP-84 SW0846+0544 2011 226 5676.372824 10.9244 0.0193 19 WASP-84 SW0846+0544 2011 226 5676.373264 10.9379 0.0178

The measurements for WASP-70 include both WASP-70A and WASP-70B. The uncertainties are the formal errors (i.e. they have not been rescaled). This table is available in its entirety via the CDS.

Table 3. Radial velocity measurements

Set Star Spectrograph BJD(UTC) RV σRV BS 2450000 − 1 1 1 (day) (km s− ) (km s− ) (km s− ) 1 WASP-69 CORALIE 5070.719440 9.61358 0.00369 0.02674 − − 1 WASP-69 CORALIE 5335.854399 9.64619 0.00376 0.01681 − − ... 2 WASP-70A CORALIE 5038.745436 65.43271 0.00840 0.03410 − − 2 WASP-70A CORALIE 5098.630404 65.48071 0.01411 0.01614 − − ... 4 WASP-84 CORALIE 6003.546822 11.52073 0.00822 0.03026 − 4 WASP-84 CORALIE 6004.564579 11.56112 0.00661 0.00007 − −

The presented RVs have not been pre-whitened and the uncertainties are the formal errors (i.e. with no added jitter). The uncertainty on bisector span (BS) is 2 σRV. This table is available in its entirety via the CDS.

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 5

Table 4. EulerCam, TRAPPIST and RISE photometry

Set Star Imager Filter BJD(UTC) Rel. flux, F σF 2450000 − (day)

1 WASP-69 EulerCam Gunn-r 5845.489188 1.000769 0.000842 1 WASP-69 EulerCam Gunn-r 5845.489624 0.999093 0.000840 ... 4 WASP-70A TRAPPIST I + z′ 5825.48554 1.006661 0.002919 4 WASP-70A TRAPPIST I + z′ 5825.48576 1.001043 0.002903 ... 8 WASP-84 RISE V + R 6311.743556 0.999610 0.001827 8 WASP-84 RISE V + R 6311.743649 1.000331 0.001829

The flux values are differential and normalised to the out-of-transit levels. The contamination of the WASP-70A photometry by WASP-70B has been accounted for. The uncertainties are the formal errors (i.e. they have not been rescaled). This table is available in its entirety via the CDS.

Table 5. Stellar parameters from spectra

Parameter WASP-69 WASP-70A WASP-70B WASP-84

T ff/ K 4700 50 5700 80 4900 200 5300 100 e ± ± ± ± log g 4.5 0.15 4.26 0.09 4.5 0.2 4.4 0.1 ∗ 1 ± ± ± ± ξt /km s 0.7 0.2 1.1 0.1 0.7 0.2 1.0 0.1 − ± ± ± ± v sin I /km s 1 2.2 0.4 1.8 0.4 3.9 0.8a 4.1 0.3 − ± ± ± ± log R 4.54 5.23 4.43 HK′ − − − [Fe/H] 0.15 0.08 0.01 0.06 0.00 0.10 ± − ± ± [Mg/H] . . . 0.05 0.04 ... ± [Si/H] 0.42 0.12 0.15 0.05 0.06 0.07 ± ± ± [Ca/H] 0.15 0.09 0.07 0.10 0.15 0.09 ± ± ± [Sc/H] 0.25 0.10 0.15 0.11 0.01 0.14 ± ± − ± [Ti/H] 0.21 0.06 0.07 0.04 0.09 0.11 ± ± ± [V/H] 0.48 0.15 0.02 0.08 0.21 0.12 ± ± ± [Cr/H] 0.18 0.18 0.05 0.04 0.08 0.15 ± ± ± [Mn/H] 0.40 0.07 ...... ± [Co/H] 0.42 0.06 0.10 0.05 0.04 0.04 ± ± ± [Ni/H] 0.29 0.11 0.05 0.05 0.02 0.06 ± ± − ± log A(Li) < 0.05 0.07 < 1.20 0.07 < 0.71 0.25 < 0.12 0.11 ± ± ± ± Sp. Typeb K5 G4 K3 K0 Agec / Gyr 2 9–10 1 ∼ ∼ Distance / pc 50 10 pc 245 20 125 20 ± ± ± Aquarius Hydra R.A. (J2000)d 21h00m06s.19 21h01m54s.48f 08h44m25s.71 Dec. (J2000)d 05 05 40. 1 13 25 59. 8f +01 51 36. 0 − ◦ ′ ′′ − ◦ ′ ′′ ◦ ′ ′′ Bd 10.93 0.06 11.75 0.13f 11.64 0.10 ± ± ± Vd 9.87 0.03 10.79 0.08f 10.83 0.08 ± ± ± Je 8.03 0.02 10.00 0.03f 9.35 0.03 ± ± ± He 7.54 0.02 9.71 0.04f 8.96 0.02 ± ± ± Ke 7.46 0.02 9.58 0.03f 8.86 0.02 ± ± ± 2MASS Je 21000618 0505398 21015446 1325595f 08442570+0151361 − − a Due to low S/N, the v sin I value for WASP-70B should be considered an upper limit. b The spectral types were estimated from Teff using the table in Gray (2008). c See Sections 6, 7 and 8. d From the Tycho-2 catalogue (Høg et al. 2000). e From the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006). f WASP-70A+B appear as a single entry in the Tycho-2 and 2MASS catalogues.

c 2013 RAS, MNRAS 000, 1–17 6 D. R. Anderson et al.

1.02 1.02 1 1 0.98 Relative flux Relative flux 0.98 0.96 0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6 1.02

1 1

WASP WASP 0.98 0.98

EulerCam 0.96 Relative flux TRAPPIST

0.96

Relative flux 0.94 TRAPPIST RISE 0.92 0.94 0.96 0.98 1 1.02 1.04

100 CORALIE 0.9 HARPS RISE -1 50 0.88 0.99 1 1.01 100

0 -1 50 -50 Relative radial velocity / m s

0 -100

0.4 0.6 0.8 1 1.2 1.4 1.6

-1 100 -50 Relative radial velocity / m s

0

-100

Bis. span / m s -100 0.4 0.6 0.8 1 1.2 1.4 1.6 100

0.4 0.6 0.8 1 1.2 1.4 1.6 -1 Orbital phase

Figure 2. WASP-70Ab discovery data. Caption as for Figure 1. Two 0 CORALIE RVs, taken during transit and depicted with open circles, were excluded from the fit as we did not model the Rossiter-McLaughlin effect. The RVs were not pre-whitened as we did not find WASP-70A to be active. Bis. span / m s -100 0.4 0.6 0.8 1 1.2 1.4 1.6 Orbital phase

Figure 3. WASP-84b discovery data. Caption as for Figure 1.

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 7

We found no evidence of modulation above the 1-mmag level Teff and [Fe/H] are proposal parameters constrained by in the WASP-70A+B lightcurves. The results for WASP-69 and Gaussian priors with mean values and variances derived directly WASP-84 are shown in Table 6. Taking the average of the peri- from the stellar spectra (see Section 3). ods for each star gives our best estimates for the rotation periods As the planet-to-star area ratio is determined from the mea- of 23.07 0.16 d for WASP-69 and 14.36 0.35 d for WASP-84. sured transit depth, the planet radius RP follows from R . The planet ± ± ∗ The WASP-69 data sets obtained in 2008 with both cameras 224 mass MP is calculated from the the measured value of K1 and the and 226 each show periods of 23/2 days, presumably as a result value of M ; the planetary density ρP and surface gravity log gP then ∼ ∗ of multiple spot groups on the surface of the star during this ob- follow. We calculate the planetary equilibrium temperature Teql, as- serving season. The data on WASP-69 obtained by camera 223 in suming zero albedo and efficient redistribution of heat from the 2008 are affected by systematic instrumental noise and so are not planet’s presumed permanent day side to its night side. We also useful for this analysis. calculate the durations of transit ingress (T12) and egress (T34). For both WASP-69 and WASP-84, we used a least-squares fit At each step in the MCMC procedure, model transit of the sinusoidal function and its first harmonic to model the ro- lightcurves and radial velocity curves are computed from the pro- tational modulation in the lightcurves for each camera and season posal parameter values, which are perturbed from the previous val- with rotation periods fixed at the best estimates. The results are ues by a small, random amount. The χ2 statistic is used to judge the shown in Figs. 6 and 7. We then subtracted this harmonic-series goodness of fit of these models to the data and the decision as to fit from the original WASP lightcurves prior to our analysis of the whether to accept a step is made via the Metropolis-Hastings rule transit. (Collier Cameron et al. 2007): a step is accepted if χ2 is lower than for the previous step and a step with higher χ2 is accepted with a probability proportional to exp( ∆χ2/2). This gives the procedure some robustness against local minima− and results in a thorough ex- 5 SYSTEM PARAMETERS FROM COMBINED ploration of the parameter space around the best-fitting solution. ANALYSES For WASP-69 and WASP-84 we used WASP lightcurves de- We determined the parameters of each system from a simulta- trended for rotational modulation (Section 4). We excluded the neous fit to all photometric and radial-velocity data. The fit was WASP-69 lightcurves from camera 223 as they suffer from greater performed using the current version of the Markov-chain Monte scatter than the lightcurves from the other cameras. The WASP- Carlo (MCMC) code described by Collier Cameron et al. (2007) 70A lightcurves were corrected for dilution by WASP-70B using and Pollacco et al. (2008). flux ratios derived from in-focus EulerCam and TRAPPIST images The transit lightcurves are modelled using the formulation (Section 6). To give proper weighting to each photometry data set, of Mandel & Agol (2002) with the assumption that the planet is the uncertainties were scaled at the start of the MCMC so as to much smaller than the star. Limb-darkening was accounted for us- obtain a photometric reduced-χ2 of unity. ing a four-coefficient, nonlinear limb-darkening model, using coef- For WASP-69b and WASP-70Ab the improvement in the fit to ficients appropriate to the passbands from the tabulations of Claret the RV data resulting from the use of an eccentric orbit model is (2000, 2004). The coefficients are interpolated once using the val- small and is consistent with the underlying orbit being circular. We ues of log g and [Fe/H] of Table 5, but are interpolated at each thus adopt circular orbits, which Anderson et al. (2012) suggest is ∗ MCMC step using the latest value of T ff. The coefficient values the prudent choice for short-period, Jupiter-mass planets in the ab- e ∼ corresponding to the best-fitting value of Teff are given in Table 7. sence of evidence to the contrary. In a far wider orbit, closer to that The transit lightcurve is parameterised by the epoch of mid-transit of the eccentric WASP-8b (Queloz et al. 2010), WASP-84b will ex- 2 T0, the orbital period P, the planet-to-star area ratio (RP/R ) , the perience weaker tidal forces and there is indication that the system ∗ approximate duration of the transit from initial to final contact T14, is young (Section 8), so there is less reason to expect its orbit to and the impact parameter b = a cos i/R (the distance, in fractional be circular. Using the F-test approach of Lucy & Sweeney (1971), ∗ stellar radii, of the transit chord from the star’s centre in the case of we calculate a 60 per cent probability that the improvement in the a circular orbit), where a is the semimajor axis and i is the inclina- fit could have arisen by chance if the underlying orbit were circu- tion of the orbital plane with respect to the sky plane. lar, which is similar to the probabilities for the other two systems. The eccentric Keplerian radial-velocity orbit is parameterised There is very little difference between the circular solution and the by the stellar reflex velocity semi-amplitude K1, the systemic veloc- eccentric solution: the stellar density and the inferred stellar and ity γ, an instrumental offset between the HARPS and CORALIE planetary dimensions differ by less than half a 1-σ error bar. Thus spectrographs ∆γHARPS, and √e cos ω and √e sin ω where e is we adopt a circular orbit for WASP-84b. We find 2-σ upper limits orbital eccentricity and ω is the argument of periastron. We use on e of 0.10, 0.067 and 0.077 for, respectively, WASP-69b, -70Ab √e cos ω and √e sin ω as they impose a uniform prior on e, and -84b. whereas the jump parameters we used previously, e cos ω and From initial MCMC runs we noted excess scatter in the RV e sin ω, impose a linear prior that biases e toward higher values residuals of both WASP-69 and WASP-84, with the residuals vary- (Anderson et al. 2011). ing sinusoidally when phased on the stellar rotation periods, as de- The linear scale of the system depends on the orbital sepa- rived from the WASP photometry (Figure 8). The RV residuals of ration a which, through Kepler’s third law, depends on the stellar WASP-69 are adequately fit by a sinusoid and those of WASP-84 mass M . At each step in the Markov chain, the latest values of benefit from the addition of the first harmonic: ∗ ρ , Teff and [Fe/H] are input in to the empirical mass calibration of ∗ RV = a sin 2πφ + b cos 2πφ + a sin 4πφ. (1) Enoch et al. (2010) (itself based on Torres, Andersen & Gim´enez activity 1 1 2 2010 and updated by Southworth 2011) to obtain M . The shapes of We phased the RV residuals of WASP-69, which were obtained the transit lightcurves and the radial-velocity curve cons∗ train stellar over two seasons, with the average (23.07 d) de- density ρ (Seager & Mall´en-Ornelas 2003), which combines with rived from all available WASP photometry. For WASP-84 we chose M to give∗ the stellar radius R . The stellar effective temperature to phase the RV residuals, which were obtained in a single season, ∗ ∗

c 2013 RAS, MNRAS 000, 1–17 8 D. R. Anderson et al.

Figure 4. Left panels: Periodograms for the WASP data from two different observing seasons for WASP-69. Horizontal lines indicate false alarm probability levels FAP=0.1,0.01,0.001. The WASP camera and of observation are noted in the titles; the first digit of the camera number denotes the observatory (1=SuperWASP-North and 2=WASP-South) and the third digit denotes the camera number. Right panels Lightcurves folded on the best periods as noted in the title.

Table 6. Frequency analysis of WASP lightcurves. Obs denotes WASP-South (S) or SuperWASP-North (N), Cam is the WASP camera number, Npoints is the number of observations, P is the period of corresponding to the strongest peak in the periodogram, Amp is the amplitude of the best-fitting sine wave in mmag and FAP is the false-alarm probability.

Year Obs Cam Baseline Npoints P/d Amp FAP

WASP-69: 2008 S 223 2008 Jun 04–Oct 12 6732 1.05 10 2.2 10 2 × − 2008 S 224 2008 Jun 04–Oct 12 6003 11.43 9 9.9 10 4 × − 2008 N 146 2008 Jun 26–Oct 08 2928 11.38 9 8.1 10 4 × − 2009 S 223 2009 May 19–Oct 11 4610 23.54 13 9.9 10 3 × − 2009 S 224 2009 May 19–Oct 11 5111 22.57 7 4.9 10 4 × − 2009 N 141 2009 Jun 26–Oct 08 6556 23.54 7 7.3 10 5 × − 2009 N 146 2009 Jun 26–Oct 08 6677 23.74 7 4.2 10 6 × − 2010 N 141 2010 Jun 26–Oct 06 6846 22.76 12 9.2 10 10 × − 2010 N 146 2010 Jun 26–Oct 06 6820 22.76 11 7.7 10 6 × − WASP-70: No evidence of modulation. WASP-84: 2009 S 226 2009 Jan 14–Apr 21 5161 13.450 10 2.1 10 9 × − 2010 N 143 2009 Dec 02–2010 Mar 31 3738 15.170 7 1.3 10 4 × − 2010 S 226 2010 Jan 15–Apr 21 4548 15.170 7 3.7 10 6 × − 2011 N 143 2010 Dec 02–2011 Mar 29 3701 14.000 15 1.8 10 12 × − 2011 S 226 2011 Jan 05–Mar 01 3684 14.000 14 1.4 10 9 × −

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 9

Figure 5. Periodograms of the WASP data for WASP-84 from four independent data sets. The WASP camera and year of observation are noted in the titles; the first digit of the camera number denotes the observatory (1=SuperWASP-North and 2=WASP-South) and the third digit denotes the camera number. Horizontal lines indicate false alarm probability levels FAP=0.1,0.01,0.001.

Table 7. Limb-darkening coefficients

Star Imager Observation bands Claret band a1 a2 a3 a4 WASP-69 WASP / EulerCam Broad (400–700 nm) / Gunn r Cousins R 0.755 0.869 1.581 0.633 − − WASP-69 TRAPPIST Sloan z Sloan z 0.824 0.922 1.362 0.549 ′ ′ − − WASP-70A WASP / EulerCam Broad (400–700 nm) / Gunn r Cousins R 0.616 0.218 0.743 0.397 − − WASP-70A TRAPPIST Cousins I + Sloan z Sloan z 0.690 0.484 0.828 0.402 ′ ′ − − WASP-84 WASP / RISE Broad (400–700 nm) / V + R Cousins R 0.695 0.591 1.269 0.589 − − WASP-84 TRAPPIST Sloan z Sloan z 0.758 0.735 1.162 0.517 ′ ′ − − with the period derived from the photometry obtained in the pre- with v sin I, but bisector span goes as (v sin I)3.3 (Saar & Donahue ceding season (2011; P = 14.0 d), as we did not perform simulta- 1997; Santos et al. 2003). The anticorrelation (r = 0.71) between rot − neous monitoring. Pre-whitening the RVs using Equation 1 reduced the RV residuals and the bisector spans of WASP-84 (v sin I = 1 the RMS of the residual scatter about the best-fitting Keplerian orbit 4.1 km s− ) is strong and the RMS of the residual RVs is reduced 1 1 1 1 from 11.04 m s− to 8.28 m s− in the case of WASP-69 and from from 14.81 m s− to 11.08 m s− (see Figure 9). The reduction in 1 1 14.81 m s− to 6.98 m s− in the case of WASP-84 (Figure 8). The the scatter of the residual RVs obtained using this approach was coefficients of the best-fitting activity models are given in Table 8. less for both WASP-69 and WASP-84 than that obtained using the harmonic-fitting approach; we thus elected to use the RVs pre- A common alternative approach is to pre-whiten RVs using a whitened by subtracting harmonic functions in the MCMC anal- relation derived from a linear fit between the RV residuals and the yses. bisector spans, which are anticorrelated when stellar activity is the source of the signal (Melo et al. 2007). The WASP-69 RV residuals For both WASP-69 and WASP-84 the best-fitting solution is are essentially uncorrelated with the bisector spans (r = 0.14) essentially unchanged by the pre-whitening. One parameter that − resulting in a small reduction in the RMS of the residual RVs, we could expect to be affected, via a modified K1, is the planetary 1 1 from 11.04 m s− to 10.08 m s− (see Figure 9). This is probably mass. For WASP-69b we obtain MP = 0.260 0.017 MJup when 1 ± due to the star’s low v sin I (2.2 km s− ) as RV varies linearly using pre-whitened RVs and M = 0.253 0.022 M otherwise. P ± Jup c 2013 RAS, MNRAS 000, 1–17 10 D. R. Anderson et al.

Figure 6. WASP lightcurves of WASP-69 plotted as a function of rota- Figure 7. WASP lightcurves of WASP-84 plotted as a function of rotation tion phase for Prot = 23.07 d. The data, detrended by SYSREM, for each phase for Prot = 14.36 d. Caption as for Fig 6. combination of observing season and camera are offset by multiples of 0.1 magnitudes. Each lightcurve is labelled with its camera number and the date range it spans, in truncated Julian Date: JD 2 450 000. Solid lines − show the harmonic fit used to remove the rotational modulation prior to modeling the transit.

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 11

K value of 0.57 suggest an age closer to 3 Gyr if we apply P √t Table 8. Coefficients of the activity-induced RV variations model ≃ using the Coma Berenices cluster colour-rotation distribution as a benchmark (Collier Cameron et al. 2009). This is consistent with Star a1 b1 a2 1 1 1 (ms− ) (ms− ) (ms− ) the lack of lithium, but slightly at odds with the log RHK′ and the Barnes (2007) gyrochronological ages. WASP-69 8.8 2.6 4.5 2.7 0 ± ± As a low-density planet in a short orbit around a relatively- WASP-84 12.1 2.3 6.5 2.8 10.9 2.5 ± − ± − ± young, active star, WASP-69b could be undergoing significant mass loss due to X-ray-driven or extreme-ultraviolet-driven evapora- tion (e.g. Lecavelier Des Etangs 2007; Jackson, Davis & Wheatley For WASP-84b we obtain MP = 0.694 0.028 MJup when using = +0.061 ± 2012). At earlier times the stellar X-ray , and hence the pre-whitened RVs and MP 0.691 0.032 MJup otherwise. To obtain a spectroscopic reduced− χ2 of unity we added a ‘jit- planetary mass loss rate, would have been higher. ROSAT recorded an X-ray source 1RXS J210010.1 050527 with a count rate of ter’ term in quadrature to the formal RV errors. Pre-whitening re- 1 − 1 1 2.4 0.9 s− over 0.1–2.4 keV and at an angular distance of 60 27′′ duced the jitter required from 10.1 m s− to6.7 ms− for WASP-69 ± ± 1 from WASP-69. There is no optical source spatially coincident with and from 13.5 m s− to zero for WASP-84; the jitter for WASP-70A 1 the ROSAT source position. The brightest object in the NOMAD was 3.9 m s− . We excluded two in-transit WASP-70A RVs from the analysis as we did not model the Rossiter-McLaughlin effect catalog located within 60′′ of the ROSAT source is 30′′ away and (e.g. Brown et al. 2012). has R=16.6, whereas WASP-69 is at 60′′ and has R=9.2. Assuming a uniform sky distribution for the 18 811 bright sources (count rate We explored the possible impact of spots on our determina- 1 tion of the system parameters of WASP-69 and WASP-84. We as- > 0.05 s− ) of the ROSAT all-sky catalogue (Voges et al. 1999), the sumed that the stars’ visible faces were made permanently dimmer, probability that an unrelated source is within 120′′ of WASP-69 is by spots located outside of the transit chords, at the level of the 0.05 per cent. maximum amplitudes of the rotational-modulation signals found Thus, assuming the ROSAT source to be WASP-69, we follow Jackson, Davis & Wheatley (2012) to estimate a cur- from the WASP photometry (Table 6). Having corrected the tran- 12 1 sit lightcurves for this, we performed MCMC analyses again and rent planetary mass loss rate of 10 g s− , having assumed found the stellar and planetary dimensions (density, radius, mass) an evaporation efficiency factor of 0.25 and having converted the count rate to an X-ray luminosity (log(L /L ) = 4.43; to have changed by less than 0.2 σ. X bol − The system parameters from the combined analyses are given Fleming, Schmitt & Giampapa 1995). This can be compared in Table 9 and the best fits to the radial velocities and the photom- with the mass-loss rates of HD 209458b and HD 189733b, 10–11 1 estimated at 10 g s− from observations of Lyman α ab- etry are plotted in Figures 1, 2 and 3. ∼ sorption by escaping hydrogen atoms (Vidal-Madjar et al. 2003; Lecavelier Des Etangs et al. 2010). Jackson, Davis & Wheatley (2012) estimate that a star of similar spectral type as WASP-69 has 6 THE WASP-69 SYSTEM a saturated X-ray luminosity ratio of log(L /L ) 3.35 for X bol sat ∼ − WASP-69b is a bloated Saturn-mass planet (0.26 M , 1.06 R ) the first 200 Myr of its life. Assuming WASP-69b to have been Jup Jup ∼ in a 3.868-d orbit around an active mid-K dwarf. The system is in-situ during this period suggests the planet would have lost mass 13 1 expected to be a favourable target for transmission spectroscopy at a rate of 10 g s− and, assuming a constant planetary density, ∼ owing to the large predicted scale height of the planet’s atmosphere the planet to have undergone a fractional mass loss of 0.2. ∼ and the apparent brightness and the small size of the star. The pa- rameters derived from the spectral analysis and the MCMC analy- sis are given, respectively, in Tables 5 and 9 and the corresponding 7 THE WASP-70 SYSTEM transit and Keplerian orbit models are superimposed, respectively, on the radial velocities and the photometry in Figure 1. WASP-70Ab is a 0.59MJup planet in a 3.713-d orbit around the pri- We estimated the stellar rotation period Prot from activity- mary of a spatially-resolved G4+K3 binary. The parameters derived rotation induced modulation of the WASP lightcurves to be 23.07 from the spectral analysis and the MCMC analysis are given, re- 0.16 d. Together with our estimate of the stellar radius (Table 9),± spectively, in Tables 5 and 9 and the corresponding transit and Ke- 1 this implies a rotation speed of v = 1.78 0.06 km s− , which can plerian orbit models are superimposed, respectively, on the radial ± be compared with the spectroscopic estimate of the projected rota- velocities and the photometry in Figure 2. 1 tion speed of v sin I = 2.2 0.4 km s− . 2MASS images reveal that the two stars are separated by 3′′ ± ∼ The rotation rate (P 6 18.7 3.5 d) implied by the v sin I gives with a position angle of 167◦. We obtained in-focus TRAPPIST ± a gyrochronological age of 0.73 0.28 Gyr using the Barnes (2007) I + z′ and EulerCam Gunn r images in 2011 and 2012, respectively ± relation. The lightcurve-modulation period implies a slightly older (Figure 11). These show the two stars separated by 3′′. 3 at a posi- gyrochronological age of 1.10 0.15 Gyr. tion angle of 167◦, which is consistent with no significant relative ± There is no significant detection of lithium in the spectra of motion since the 2MASS images were taken in 1998. The projected WASP-69, with an equivalent-width upper limit of 12 m, corre- separation of 3′′. 3 and inferred distance of 245 pc, give a physical sponding to an abundance upper limit of log A(Li) < 0.05 0.07. separation of the two stars of at least 800 AU. Further evidence of ± This implies an age of at least 0.5 Gyr (Sestito & Randich 2005). the stars’ association comes from the mean radial velocities mea- 1 1 There are strong emission peaks evident in the Ca II H+K sured from CORALIE spectra: 65.4 km s− and 64.6 km s− for − − lines (Figure 10), with an estimated activity index of log RHK′ WASP-70A and WASP-70B, respectively. 4.54. This gives an approximate age of 0.8 Gyr according to∼ We determined the flux ratios of the stellar pair from aper- −Mamajek & Hillenbrand (2008), which is consistent∼ with the age ture photometry of the images. For the EulerCam images, we used implied from the rotation rate and the absence of lithium. the USNO-B1.0 magnitudes of comparison stars to determine ap- Interestingly, the observed rotation period and the 2MASS J proximate r-band magnitudes (Table 10). By combining these with − c 2013 RAS, MNRAS 000, 1–17 12 D. R. Anderson et al.

Figure 8. The radial-velocity modulation induced by stellar activity. The panels on the left pertain to WASP-69 and those on the right pertain to WASP-84; the remainder of the caption describes the plots of both stars. (a): The residuals about the best-fitting Keplerian orbit phased on a period derived from a modulation analysis of the WASP photometry. The best-fitting harmonic function is overplotted (see Equation 1 and Table 8). (b): The residuals about the best-fitting Keplerian orbit as a function of time; the best-fitting harmonic function is overplotted. The abscissa scales on the two adjacent panels are equal. Specific to WASP-69, two points are not plotted; one was taken a season earlier and the other a season later than the data shown. With reference to the top-left panel (panel (a) for WASP-69), the ‘early’ point is at coordinate (0.61, 19) and the ‘late’ point is at coordinate (0.51, 12). (c): The RVs, both detrended and non-detrended, − − folded on the transit ephemeris. By first detrending the RVs with the harmonic function a significantly lower scatter is obtained (the blue circles about the dashed line) as compared to the non-detrended RVs (the red squares about the solid line).

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 13

Figure 9. The anticorrelated variations in bisector span and radial velocity induced by stellar activity. From initial MCMCs for both WASP-69 and WASP-84, we noted large scatter about the best-fitting Keplerian orbits, which we suggest to be the product of stellar activity. The panels on the left pertain to WASP-69 and those on the right pertain to WASP-84; the remainder of the caption describes the plots of both stars. (a): Similar to Melo et al. (2007), we found the residuals to be anticorrelated with bisector span. A linear fit to the data is overplotted and the correlation coefficient, r is given. (b): By first detrending the RVs with this bisector-residual relation, a smaller scatter about the best-fitting Keplerian orbit is obtained (the blue circles about the dashed line) as compared to the non-detrended RVs (the red squares about the solid line). The result is inferior to that obtained when detrending the RVs with low-order harmonic series (Figure 8).

the Teff from the spectral analysis we constructed a Hertzsprung- an age of at least 5 Gyr (Sestito & Randich 2005) for WASP-70A Russell diagram for the WASP-70 double system (Figure 12). A and over 600 Myr∼ for WASP-70B. We found no evidence of modu- distance modulus of 6.95 0.15 ( 245 20 pc) was required lation in the WASP-70A+B lightcurves or of emission peaks in the ± ≡ ± to bring the companion star on to the . WASP-70A Ca II H+K lines in either star. appears to have evolved off the ZAMS with an age of around 9– 10 Gyr. 8 THE WASP-84 SYSTEM The rotation rate for WASP-70A (P 6 34.1 7.8 d) implied by ± the v sin I gives a gyrochronological age of 6 7 3 Gyr using the WASP-84b is a 0.69-MJup planet in an 8.523-d orbit around an Barnes (2007) relation. There is no significant detection± of lithium early-K dwarf. After HAT-P-15b and HAT-P-17b, which have or- in the spectra, with equivalent-width upper limits of 11m and 20m, bital periods of 10.86 d and 10.34 d, respectively, WASP-84b has corresponding to abundance upper limits of log A(Li) < 1.20 0.07 the longest orbital period of the planets discovered from the ground and < 0.71 0.25 for WASP-70A and B, respectively. These± imply by the transit technique (Kov´acs et al. 2010; Howard et al. 2012). ± c 2013 RAS, MNRAS 000, 1–17 14 D. R. Anderson et al.

Table 9. System parameters

Parameter Symbol (Unit) WASP-69b WASP-70Ab WASP-84b

Orbital period P (d) 3.8681382 0.0000017 3.7130203 0.0000047 8.5234865 0.0000070 ± ± ± Epoch of mid-transit Tc (BJD, UTC) 2455748.83344 0.00018 2455736.50348 0.00027 2456286.10583 0.00009 ± +0.0014± ± Transit duration T14 (d) 0.0929 0.0012 0.1389 0.0019 0.11452 0.00046 = ± +−0.0016 ± Transit ingress/egress duration T12 T34 (d) 0.0192 0.0014 0.0150 0.0021 0.02064 0.00060 ∆ = 2 2 ± −+0.00022 ± Planet-to-star area ratio F RP/R 0.01786 0.00042 0.00970 0.00026 0.01678 0.00015 ∗ ± +−0.076 ± Impact parameter b 0.686 0.023 0.431 0.169 0.632 0.012 ± −+1.24 ± Orbital inclination i (◦) 86.71 0.20 87.12 0.65 88.368 0.050 ± − ± 1 Stellar reflex velocity semi-amplitude K1 (ms− ) 38.1 2.4 72.3 2.0 77.4 2.0 1 ± ± ± Systemic velocity γ (ms− ) 9 628.26 0.23 65 389.74 0.32 11 578.14 0.33 1 − ± − ± − ± Offset between CORALIE and HARPS ∆γHARPS (ms ) — 15.44 0.11 — − ± Eccentricity e 0 (adopted) (<0.10 at 2 σ) 0 (adopted) (<0.067 at 2 σ) 0 (adopted) (<0.077 at 2 σ) Stellar mass M (M ) 0.826 0.029 1.106 0.042 0.842 0.037 ∗ ⊙ ± ±+ ± Stellar radius R (R ) 0.813 0.028 1.215 0.064 0.748 0.015 ∗ ⊙ ± +0.089 ± Stellar surface gravity log g (cgs) 4.535 0.023 4.314−0.052 4.616 0.011 ∗ ± +0.036 ± Stellar density ρ (ρ ) 1.54 0.13 0.619−0.136 2.015 0.070 ∗ ⊙ ± 0.077 ± Stellar effective temperature T ff (K) 4715 50 5763− 79 5314 88 e ± ± ± Stellar metallicity [Fe/H] 0.144 0.077 0.006 0.063 0.00 0.10 ± − ± ± Planetary mass MP (MJup) 0.260 0.017 0.590 0.022 0.694 0.028 ± ±+0.073 ± Planetary radius RP (RJup) 1.057 0.047 1.164 0.102 0.942 0.022 ± −+0.066 ± Planetary surface gravity log gP (cgs) 2.726 0.046 3.000 0.050 3.253 0.018 ± −+0.104 ± Planetary density ρP (ρJ) 0.219 0.031 0.375 0.830 0.048 ± 0.060 ± Orbital major semi-axis a (AU) 0.04525 0.00053 0.04853 −0.00062 0.0771 0.0012 ± ± ± Planetary equilibrium temperature T (K) 963 18 1387 40 797 16 eql ± ± ±

0.09

0.08

0.07

0.06

0.05

0.04 Relative Flux 0.03

0.02

0.01

0 3880 3900 3920 3940 3960 3980 4000 4020 Wavelength (Å)

Figure 10. CORALIE spectrum of WASP-69 showing strong emission in the Ca II H+K line cores.

Table 10. Estimated magnitudes and flux ratios of WASP-70A+B

Band WASP-70A WASP-70B ∆A B fB/ fA − Figure 11. Upper portion: EulerCam Gunn-r image of WASP-70A+B; the r 10.8 13.1 2.3 0.12 − B component is at a distance of 3′′. 3 and position angle of 167◦. Lower I + z ...... 1.6 0.22 ′ − portion: The counts per pixel along the slice marked on the image.

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 15

1.06 RJup) in a 3.868-d period around a mid-K dwarf. We find

3 WASP-69 to be active from emission peaks in the Ca II H+K lines (log RHK′ 4.54) and from modulation of the star’s lightcurves (P = 23∼.07 − 0.16 d; amplitude = 7–13 mmag). Both the gy- rot ± 4 rochronological calibration of Barnes (2007) and the age-activity relation of Mamajek & Hillenbrand (2008) suggest an age of

5 around 1 Gyr, whereas the Coma Berenices colour-rotation cali-

(absolute mag) bration of (Collier Cameron et al. 2009) suggests an age closer to r 3 Gyr.

Gunn 6 WASP-70Ab is a Jupiter-type planet (0.59 MJup, 1.16 RJup) in a 3.713-d orbit around the primary of a spatially-resolved G4+K3 7 binary separated by 3′′. 3(>800 AU). An absence of emission in the Ca II H+K lines and an absence of lightcurve modulation indicates 3.8 3.78 3.76 3.74 3.72 3.7 3.68 3.66 3.64

log Teff WASP-70A is relatively inactive. We used the binary nature of the system to construct an H-R diagram, from which we estimate its Figure 12. H-R diagram for WASP-70A+B. Various isochrones from age to be 9–10 Gyr. Marigo et al. (2008) are plotted, with ages of 0, 3, 5, 9 and 10 Gyr. WASP-84b is a Jupiter-type planet (0.69 MJup, 0.94 RJup) in an 8.523-d orbit around an active early-K dwarf. The planet has the The parameters derived from the spectral analysis and the MCMC third-longest period of the transiting planets discovered from the analysis are given, respectively, in Tables 5 and 9 and the corre- ground. We find WASP-84 to be active from emission peaks in the sponding transit and Keplerian orbit models are superimposed, re- Ca II H+K lines (log R′ 4.43) and from modulation of the HK ∼ − spectively, on the radial velocities and the photometry in Figure 3. star’s lightcurves (P = 14.36 0.35 d; amplitude = 7–15 mmag). ± We estimated the stellar rotation period from the modulation The gyrochronological calibration of Barnes (2007) suggests an of the WASP lightcurves to be P = 14.36 0.35 d. Together with age of 0.8 Gyr for WASP-84, whereas the age-activity relation of rot ± ∼ our estimates of the stellar radius (Table 9), this implies a rota- Mamajek & Hillenbrand (2008) suggests an age of 0.4 Gyr. ∼ tion speed of v = 2.64 0.08 km s 1. This is inconsistent with the − We used the empirical relations of spectroscopic estimate± of the projected rotation speed of v sin I= Enoch, Collier Cameron & Horne (2012), derived from fits to 4.1 0.3 km s 1. This disagreement may be due to an overesti- − the properties of a well-characterised sample of transiting planets, mation± of v sin I and an underestimation of R . We took macrotur- ∗ to predict the radii of the three presented planets. The difference bulence values from the tabulation of Bruntt et al. (2010a). If we between the observed radii (Table 9) and the predicted radii of instead adopted a higher macroturbulence value from Gray (2008) ∆ 1 both WASP-70Ab ( RP,obs pred = 0.067 RJup) and WASP-84b ( then we would obtain a lower v sin I of 3.5 km s− . Similarly, sys- − − ∼ (∆RP,obs pred = 0.065 RJup)) are small. For comparison, the average tematics in the lightcurves or a limitation of the empirical mass − difference between the predicted and observed radii for the sample calibration could have resulted in an underestimation of the stellar of Enoch, Collier Cameron & Horne (2012) is 0.11 R . The radius. Jup predicted radius of WASP-69b (0.791 R ) is smaller than the Using the Barnes (2007) relation, P gives a gyrochronologi- Jup rot observed radius (Table 9) by 0.266 R , showing the planet to be cal age of 0.79 0.12 Gyr, whereas the rotation rate implied by the Jup bloated. v sin I (P 6 9.2± 0.7 d) gives 60.34 0.07 Gyr. ± ± There is no significant detection of lithium in the spectra, with We observed excess scatter in the radial-velocity (RV) resid- an equivalent width upper limit of 8m, corresponding to an abun- uals about the best-fitting Keplerian orbits for the two active dance upper limit of log A(Li) < 0.12 0.11. This implies an age stars WASP-69 and WASP-84. Phasing the RV residuals on the ± of at least 0.5 Gyr (Sestito & Randich 2005). photometrically-determined stellar rotation periods showed the ex- ∼ There are emission peaks evident in the Ca II H+K lines, with cess scatter to be induced by a combination of stellar activity and an estimated activity index of log R′ = 4.43. This gives an rotation. We fit the residuals with low-order harmonic series and HK ∼ − approximate age of 0.4 Gyr according to Mamajek & Hillenbrand subtracted the best fits from the RVs prior to deriving the systems’ (2008). parameters. The systems’ solutions were essentially unchanged by Using the Coma Berenices cluster colour-rotation relation of this, with much less than a 1-σ change to the planet mass in each Collier Cameron et al. (2009) we find an age of 1.4 Gyr. As was the case. We found this method of pre-whitening using a harmonic fit case with WASP-69, this is at odds with the log RHK′ and the Barnes to result in a greater reduction in the residual RV scatter (∆RMS 1 1 (2007) gyrochronological ages. = 2.76 m s− for WASP-69 and ∆RMS = 7.83 m s− for WASP- 84)− than the more traditional method of pre-whitening− with a fit 1 to the RV residuals and the bisector spans (∆RMS = 0.96 m s− 1 − 9 DISCUSSION AND SUMMARY for WASP-69 and ∆RMS = 3.73 m s− for WASP-84). The poor − performance of the bisector-RV fit method for WASP-69 probably We reported the discovery of the transiting exoplanets WASP-69b, stems from the slow rotation of the star, which has been shown WASP-70Ab and WASP-84b, each of which orbits a bright star to limit the method’s efficacy (Saar & Donahue 1997; Santos et al. (V 10). We derived the system parameters from a joint analysis 2003). of the∼ WASP survey photometry, the CORALIE and HARPS ra- dial velocities and the high-precision photometry from TRAPPIST, WASP-69 and WASP-84 are two of the most active stars EulerCam and RISE. known to host exoplanets. Of the 303 systems with measured

WASP-69b is a bloated Saturn-type planet (0.26 MJup, log RHK′ values only 17 are more active than WASP-69 and only

c 2013 RAS, MNRAS 000, 1–17 16 D. R. Anderson et al.

6 are more active than WASP-841. Conversely, only 14 exoplanet Faedi F. et al., 2011, Detection and Dynamics of Transiting Exo- host stars have measured log RHK′ indices indicating lower activity planets, St. Michel l’Observatoire, France, Edited by F. Bouchy; levels than WASP-70A, though values for inactive stars are proba- R. D´ıaz; C. Moutou; EPJ Web of Conferences, Volume 11, bly underreported. id.01003, 11, 1003 Ferraz-Mello S., Tadeu Dos Santos M., Beaug´eC., Michtchenko T. A., Rodr´ıguez A., 2011, A&A, 531, A161 ACKNOWLEDGEMENTS Fleming T. A., Schmitt J. H. M. M., Giampapa M. S., 1995, ApJ, 450, 401 WASP-South is hosted by the South African Astronomical Ob- Gillon M. et al., 2011, A&A, 533, A88 servatory and SuperWASP-North is hosted by the Isaac Newton —, 2009, A&A, 496, 259 Group on La Palma. We are grateful for their ongoing support and Gray D. F., 2008, The Observation and Analysis of Stellar Photo- assistance. Funding for WASP comes from consortium universi- spheres. Cambridge University Press ties and from the UK’s Science and Technology Facilities Coun- Hatzes A. P. et al., 2011, ApJ, 743, 75 cil. TRAPPIST is funded by the Belgian Fund for Scientific Re- Hellier C. et al., 2011, Detection and Dynamics of Transiting Exo- search (Fond National de la Recherche Scientifique, FNRS) under planets, St. Michel l’Observatoire, France, Edited by F. Bouchy; the grant FRFC 2.5.594.09.F, with the participation of the Swiss R. D´ıaz; C. Moutou; EPJ Web of Conferences, Volume 11, National Science Fundation (SNF). The Liverpool Telescope is op- id.01004, 11, 1004 erated on the island of La Palma by Liverpool John Moores Uni- Høg E. et al., 2000, A&A, 355, L27 versity in the Spanish Observatorio del Roque de los Muchachos Howard A. W. et al., 2012, ApJ, 749, 134 of the Instituto de Astrofisica de Canarias with financial support Hu´elamo N. et al., 2008, A&A, 489, L9 from the UK Science and Technology Facilities Council. M. Gillon Jackson A. P., Davis T. A., Wheatley P. J., 2012, MNRAS, 422, and E. Jehin are FNRS Research Associates. A. H.M.J. Triaud is 2024 a Swiss National Science Foundation fellow under grant number Jehin E. et al., 2011, The Messenger, 145, 2 PBGEP2-145594. L. Delrez is a FNRS/FRIA Doctoral Fellow. We Kov´acs G., Bakos G., Noyes R. W., 2005, MNRAS, 356, 557 are grateful to R. D. Jeffries for discussion of the ROSAT catalog. Kov´acs G. et al., 2010, ApJ, 724, 866 This research has made use of the Exoplanet Orbit Database and the Lanza A. F. et al., 2010, A&A, 520, A53 Exoplanet Data Explorer at http://exoplanets.org and the Lecavelier Des Etangs A., 2007, A&A, 461, 1185 VizieR catalogue access tool, CDS, Strasbourg, France. The origi- Lecavelier Des Etangs A. et al., 2010, A&A, 514, A72 nal description of the VizieR service was published in A&AS 143, Lendl M. et al., 2012, A&A, 544, A72 23. Lucy L. B., Sweeney M. A., 1971, AJ, 76, 544 Magain P., 1984, A&A, 134, 189 Mamajek E. E., Hillenbrand L. A., 2008, ApJ, 687, 1264 REFERENCES Mandel K., Agol E., 2002, ApJ, 580, L171 Mandell A. M., Drake Deming L., Blake G. A., Knutson H. A., Anderson D. R. et al., 2012, MNRAS, 422, 1988 Mumma M. J., Villanueva G. L., Salyk C., 2011, ApJ, 728, 18 —, 2011, ApJ, 726, L19 Marigo P., Girardi L., Bressan A., Groenewegen M. A. T., Silva Angerhausen D., Krabbe A., 2011, in Astronomical Society of the L., Granato G. L., 2008, A&A, 482, 883 Pacific Conference Series, Vol. 450, Astronomical Society of the Maxted P. F. L. et al., 2013, PASP, 125, 48 Pacific Conference Series, Beaulieu J. P., Dieters S., Tinetti G., —, 2011, PASP, 123, 547 eds., p. 55 Melo C. et al., 2007, A&A, 467, 721 Bakos G., Noyes R. W., Kov´acs G., Stanek K. Z., Sasselov D. D., Orosz J. A. et al., 2012, Science, 337, 1511 Domsa I., 2004, PASP, 116, 266 Pepe F. et al., 2005, The Messenger, 120, 22 Bakos G. A.´ et al., 2013, PASP, 125, 154 Pollacco D. et al., 2008, MNRAS, 385, 1576 Baranne A. et al., 1996, A&AS, 119, 373 Pollacco D. L. et al., 2006, PASP, 118, 1407 Barnes S. A., 2007, ApJ, 669, 1167 Pont F., Aigrain S., Zucker S., 2011, MNRAS, 411, 1953 Brown D. J. A. et al., 2012, MNRAS, 423, 1503 Queloz D. et al., 2010, A&A, 517, L1+ Bruntt H. et al., 2010a, MNRAS, 405, 1907 —, 2009, A&A, 506, 303 —, 2010b, A&A, 519, A51 —, 2001, A&A, 379, 279 Claret A., 2000, A&A, 363, 1081 Saar S. H., Donahue R. A., 1997, ApJ, 485, 319 —, 2004, A&A, 428, 1001 Santos N. C. et al., 2003, A&A, 406, 373 Collier Cameron A. et al., 2009, MNRAS, 400, 451 Seager S., Mall´en-Ornelas G., 2003, ApJ, 585, 1038 —, 2006, MNRAS, 373, 799 Sestito P., Randich S., 2005, A&A, 442, 615 —, 2007, MNRAS, 380, 1230 Skrutskie M. F. et al., 2006, AJ, 131, 1163 Desidera S. et al., 2004, A&A, 420, L27 Southworth J., 2011, MNRAS, 417, 2166 Doyle A. P. et al., 2013, MNRAS, 428, 3164 Steele I. A., Bates S. D., Gibson N., Keenan F., Meaburn J., Mot- Doyle L. R. et al., 2011, Science, 333, 1602 tram C. J., Pollacco D., Todd I., 2008, in Society of Photo- Enoch B., Collier Cameron A., Horne K., 2012, A&A, 540, A99 Optical Instrumentation Engineers (SPIE) Conference Series, Enoch B., Collier Cameron A., Parley N. R., Hebb L., 2010, A&A, Vol. 7014, Society of Photo-Optical Instrumentation Engineers 516, A33 (SPIE) Conference Series Steele I. A. et al., 2004, in Society of Photo-Optical Instrumen- 1 The data were retrieved from http://exoplanets.org on 2013 tation Engineers (SPIE) Conference Series, Vol. 5489, Ground- May 5. See Wright et al. (2011). based Telescopes, Oschmann Jr. J. M., ed., pp. 679–692

c 2013 RAS, MNRAS 000, 1–17 WASP-69b, WASP-70Ab and WASP-84b 17

Swain M. R. et al., 2010, Nature, 463, 637 Tamuz O., Mazeh T., Zucker S., 2005, MNRAS, 356, 1466 Torres G., Andersen J., Gim´enez A., 2010, A&A Rev., 18, 67 Vidal-Madjar A., Lecavelier des Etangs A., D´esert J.-M., Ballester G. E., Ferlet R., H´ebrard G., Mayor M., 2003, Nature, 422, 143 Voges W. et al., 1999, A&A, 349, 389 Wright J. T. et al., 2011, PASP, 123, 412

c 2013 RAS, MNRAS 000, 1–17